Cartan subgroup

From HandWiki

In the theory of algebraic groups, a Cartan subgroup of a connected linear algebraic group [math]\displaystyle{ G }[/math] over a (not necessarily algebraically closed) field [math]\displaystyle{ k }[/math] is the centralizer of a maximal torus. Cartan subgroups are smooth (equivalently reduced), connected and nilpotent. If [math]\displaystyle{ k }[/math] is algebraically closed, they are all conjugate to each other. [1]

Notice that in the context of algebraic groups a torus is an algebraic group [math]\displaystyle{ T }[/math] such that the base extension [math]\displaystyle{ T_{(\bar{k})} }[/math] (where [math]\displaystyle{ \bar{k} }[/math] is the algebraic closure of [math]\displaystyle{ k }[/math]) is isomorphic to the product of a finite number of copies of the [math]\displaystyle{ \mathbf{G}_m=\mathbf{GL}_1 }[/math]. Maximal such subgroups have in the theory of algebraic groups a role that is similar to that of maximal tori in the theory of Lie groups.

If [math]\displaystyle{ G }[/math] is reductive (in particular, if it is semi-simple), then a torus is maximal if and only if it is its own centraliser [2] and thus Cartan subgroups of [math]\displaystyle{ G }[/math] are precisely the maximal tori.

Example

The general linear groups [math]\displaystyle{ \mathbf{GL}_n }[/math] are reductive. The diagonal subgroup is clearly a torus (indeed a split torus, since it is product of n copies of [math]\displaystyle{ \mathbf{G}_m }[/math] already before any base extension), and it can be shown to be maximal. Since [math]\displaystyle{ \mathbf{GL}_n }[/math] is reductive, the diagonal subgroup is a Cartan subgroup.

See also

References

  1. Milne (2017), Proposition 17.44.
  2. Milne (2017), Corollary 17.84.